On the Origin of The Soft X-ray Excess in Radio Quiet AGNs. P.O. Petrucci Ursini, Cappi, Bianchi, Matt, DeRosa, Malzac, Henri

Délégation On the Origin of The Soft X-ray Excess in Radio Quiet AGNs P.O. Petrucci Ursini, Cappi, Bianchi, Matt, DeRosa, Malzac, Henri XMM the next Decade. Madrid. 9-11 May 2016

Context Soft excess: excess of flux below 1-2 kev compared to the extrapolation of the 2-10 kev power law

Context Soft excess: excess of flux below 1-2 kev compared to the extrapolation of the 2-10 kev power law Known since the 80s (Arnaud et al. 1985, Singh et al. 1985) 1068 J. Crummy et al. EXOSAT XMM XMM counts/cm 2 sec kev data Mkn 841 Arnaud et al. (1985) model expectation Ratio data/model 1 1.5 2 2.5 Mkn 841 Petrucci et al. (2007) counts/sec/kev Crummy et al. (2006) 0.1 1 10 0.5 1 Energy 2 (kev) 5 0.5 1 2 Energy (kev) Energy (kev) Energy (kev) Fig. 5. Ratio data/model for part 1 of OBS 4. The model is a simple power law fitted between 3 and 10 kev and extrapolated down to low energies. A strong soft excess and a line near 6 kev are clearly appar- Figure 1. The soft excesses of all the sources in our sample. The sources were fit as in Table 2, then the blackbody component was removed and the resi were plotted. Absorption edges were left in the model to show the form of the soft excess more clearly. The top three sources, in descending order of fl 0.5 kev, are ARK 564, NGC 4051 and TON S180; note the unusually extended shape of the TON S180 soft excess.

Context Soft excess: excess of flux below 1-2 kev compared to the extrapolation of the 2-10 kev power law Known since the 80s (Arnaud et al. 1985, Singh et al. 1985) Detected in more than 50% of Sy 1 galaxies (e.g. Halpern 1984; Turner & Pounds 1989, Piconcelli et al. 2005, Bianchi et al. 2009, Scott et al. 2012)

An explanation for the AGN soft excess 1073 Table 4. Redshifted absorption edges and narrow 6.4-keV Gaussian emission lines included in the relativistically blurred disc fits in Table 3. Energy is the energy of the absorption edge, τ is the optical depth and width is the equivalent width of the Gaussian line. Energies given without associated errors are fixed at 0.74 or 0.87 kev, the energy of the O VII and O VIII K absorption edges. Blank spaces indicate no feature is present. Context Source Energy (kev) τ 0.6+0.2 0.2 0.3+0.2 0.1 0.153+0.005 0.007 0.24+0.04 0.02 +0.01 0.10 0.01 0.37+0.03 0.03 0.40+0.02 0.01 PG 1307+085 0.74 PG 1309+355 NGC 4051 0.74 0.74 IRAS 13349+2438 0.72+0.20 0.02 0.68+0.53 0.01 0.72+0.02 0.01 0.708+0.005 0.003 Energy (kev) τ Width (ev) 0.87 0.463+0.007 0.009 0.2+0.1 0.2 0.092+0.006 0.017 90+20 20 soft excess AGN real? 3 soft Isthe soft excess inin AGN real? 33excess IsIsthe the soft excess ininagn AGN real? 3the Isthe the soft excess AGN real? soft excess Isthe the exces IsIs the in Is soft excess in real? Is 3soft the soft exc Is soft exce L9 Is the soft excess in AGN real? ARK 564 PKS 0558-504 MRK 0766 Soft excess: excess of flux below 1-2 kev compared to Bianchi et al. (2009) Gierlinski & Done (2004) Crummy et al. (2006) the extrapolation of the 2-10 kev power law 429 5.4 NGC 1.0 4051 0.2 The observation of NGC 4051 we analyse is of very high quality, and shows some features not observable in any of the other sources. It is clear from Fig. 7 that the disc reflection model follows the shape of the continuum very well, but there are several small features which contribute to the fairly high χ 2ν of 1.240 for 1040 d.o.f. Some of these features may be calibration problems, such as around the XMM Newton instrumental Au M edge at 2.3 kev, and some may be due to small amounts of absorption or emission. However, it is also0.1 possible that some of the approximations made in the disc reflection model are invalid when analysing observations of this detail (see Section 6). A more detailed analysis of NGC 4051 and its variability using the disc reflection model is in Ponti et al. (in preparation). ktbb(kev) S. Bianchi et al.: CAIXA: a catalogue of AGN In the XMM-Newton archive. I. ktbb(kev) ktbb(kev) Is the soft ex 50+40 40 Know since the 80s (Arnaud et al. 1985, Singh et al. 1985) 0.1 5.5 E 1346+266 Detected in more than 50% of Sy 1 galaxies (e.g. Halpern E 1346+266 is unusual in our sample as it has a high redshift, z = 0.915. This enables us to investigate the effectiveness of the disc 0 0.01model at source energies up to 23 kev. The simple model, reflection the disc reflection model and the version of the disc reflection model with a non-rotating black hole are all roughly equal in quality of fit to the source, so whilst the disc reflection model has proved itself to Figure 3. Soft excess blackbody temperature (kt) measured for the simple work excess at least as well as theobjects simple model this energy range it is Fig. 9. Temperature of themodel black model, used4 and to Table model in the ofincaixa. It appears completely unrelated to plottedbody against black hole masswhen (M; see Section 1). Thethe soft difficult to say anything definite about the source. The disc reflection soft excess is confined to a narrow temperature band over several decades in the X-ray luminosity (leftmass, andas earlier the noted BHbymass (right). The diﬀerent symbols refer to the classification theandobjects, model essentially works by fitting the soft of excess any excesson the basis of their absolute Gierlin ski & Done (2004) and others. iron emission. The soft is less visible in high redshift AGN 1 (no Hβ FWHM measure magnitude and Hβ FWHM: NLSY, narrow-line Seyfert 1; BLSY, broad-line Seyfert 1; excess NCSY, not-classified Seyfert due to being both shifted out of the instrument band and strongly 20 2 BLQ, broad-line quasar; NCQ, not-classified quasar (no Hβ FWHM measure available). available); NLQ, narrow-line quasar; 1.55 10 cm, from Table 1) can account for the shape of these absorbed. The disc reflection model will therefore be most useful in residuals, with a χ 2ν of 1.066 for 778 d.o.f. A simple model fit with high redshift sources which have excesses in the continuum around no Galactic absorption is still inadequate, with a χ 2ν of 1.264 for 782 6.4 kev due to iron, although it is able to fit sources where this is d.o.f. These fits are unlikely to be physical, the Galactic column is not the case. 19 2 rarely lower than 5 10 cm. The residuals are instead likely to be due to some curvature of the soft excess not addressed by our 5.6 PG 1211+143 models. PG 1211+143 has a feature in the 4 8 kev region which is not predicted by either model (see Fig. 8). The shape of the residuals to 5.3 PG 1404+226 the fits suggests two possibilities, an absorption edge at 7 kev or a broad Laor iron line. We investigated the latter possibility by adding PG 1404+226 has been previously fit with the disc reflection model a Laor line at 6.4 kev (in the source frame) to our reflection model, in Crummy et al. (2005). PG 1404+226 is strongly variable, and with the inclination, emissivity index and outer radius fixed to the when the simple model is used it appears to have an absorption same values as our Laor blurring, allowing only the inner radius feature in the low state which disappears in the high state (see and normalization to vary. We found that this improved the fit very also Dasgupta et al. 2005). Crummy et al. show that the disc resignificantly, to a χ 2ν of 1.053 for 932 d.o.f. However, this is unphysflection model can reproduce both the high-spectral states and the low-spectral states without the need for variable absorption. ical as any iron line would also have other reflected lines associated 1984; Turner & Pounds 1989, Piconcelli et al. 2005, Bianchi et al. 2009, Scott et al. 2012) With a «universal» spectral shape (e.g. Walter & Fink 1993, Gierlisnki & Done 2004, Crummy et al. 2006, Bianchi ettheal. soft2009) excess. This component is very common in CAIXA, in agreement with previous studies. We chose the ratio between the soft X-ray luminosity (rest-frame 0.5 2 kev) and the hard X-ray luminosity (rest-frame 2 10 kev) as a modelindependent indicator of the strength of the soft excess. With respect to this parameter, narrow- and broad-line objects are significantly diﬀerent populations. Indeed, it is very interesting to note that no narrow-line object is found with a value of the luminosity ratio lower than unity.

Modeling Two main models proposed yet Blurred ionized reflection e.g. Crummy et al. (2006) Thermal Comptonisation NGC 5548 e.g. Magdziarz et al. (1998) See Boissay et al. (2015) for testing/comparing both soft X-ray excess models on a large sample

Modeling Two main models proposed yet Blurred ionized reflection e.g. Crummy et al. (2006) Thermal Comptonisation NGC 5548 e.g. Magdziarz et al. (1998) «hot» " corona See Boissay et al. (2015) for testing/comparing both soft X-ray excess models on a large sample

Modeling Two main models proposed yet Blurred ionized reflection e.g. Crummy et al. (2006) Thermal Comptonisation NGC 5548 e.g. Magdziarz et al. (1998) «warm» " corona «hot» " corona See Boissay et al. (2015) for testing/comparing both soft X-ray excess models on a large sample

Thermal Comptonisation Two-phases Radiative Equilibrium Let s call Ls the soft photon luminosity entering the corona, and Lh the corona heating luminosity Corona Lh τ, kte Ls Disk Ldisk

Thermal Comptonisation Two-phases Radiative Equilibrium Let s call Ls the soft photon luminosity entering the corona, and Lh the corona heating luminosity Corona Lh τ, kte L h = L h,up + L h,down ' 2L h,up Ls Disk Ldisk

Thermal Comptonisation Two-phases Radiative Equilibrium Let s call Ls the soft photon luminosity entering the corona, and Lh the corona heating luminosity Corona Lh Ls τ, kte L h = L h,up + L h,down ' 2L h,up L s = L s,up + L s,down = L s e + L s(1 e ) 2 {z } L s,up + L s(1 e ) 2 {z } L s,down Disk Ldisk soft photons crossing the corona without being scattered

Thermal Comptonisation Two-phases Radiative Equilibrium Let s call Ls the soft photon luminosity entering the corona, and Lh the corona heating luminosity Disk Corona Lh Ls Ldisc τ, kte L h = L h,up + L h,down ' 2L h,up L s = L s,up + L s,down Constrains: = L s e + L s(1 e ) 2 {z } L s,up L obs = L s,up + L h,up + L s(1 e ) 2 {z } L s,down L s apple L disc = L h,down + L s,down +L disc,intr {z } reprocessing

Thermal Comptonisation Two-phases Radiative Equilibrium We can deduce a lower limits of the disc intrinsic emission from our best fits: Corona Lh τ, kte L disc,intr = L s b + e L obs 1 L s L disc,intr L s = 1 b + G(, L obs ) L s min Disk Ls Ldisc Examples: and b? No intrinsic disc emission and τ <<1 Lobs/Ls=2 (Haardt & Maraschi 1993) No intrinsic disc emission and τ >>1 Lobs/Ls=1

Numerical estimates of L disc,intr L s! min 1. Warm corona model: nthcomp in XSPEC. Model parameters: Γ, kte,tbb 2. Choose Tbb (e.g. 3 ev), vary Γ and kte 3. Estimate τ(γ, kte) (e.g. Beloborodov 1999) 4. Compute Lobs(Γ, kte) and Ls(Γ, kte) assuming photon conservation and deduce Ldisc,intr/Ls Tbb / E kte

Numerical estimates of L disc,intr L! s min 1. Warm corona model: nthcomp in XSPEC. Model parameters: Γ, kte,tbb 2. Choose Tbb (e.g. 3 ev), vary Γ and kte 3. Estimate τ(γ, kte) (e.g. Beloborodov 1999) 4. Compute Lobs(Γ, kte) and Ls(Γ, kte) assuming photon conservation and deduce Ldisc,intr/Ls

Sample In the CAIXA catalogue (XMM archive, Bianchi et al. 2009 + new sources) With more than 3 observations With more than 3 OM filters Low Nh (< 10 22 cm 2 ), weak WA Mkn 509, Mkn 883, Mkn 279, PG1116+215, PG1114+445, Mkn 335, NGC 7469, 1H0707-495, H0557-385, Mkn766, PG1211+143, Q2251-178

Model kev 2 (Photons cm 2 s 1 kev 1 ) 10 4 10 3 0.01 0.1 «hot» corona 10 4 10 3 0.01 0.1 1 10 100 Energy (kev)

Model kev 2 (Photons cm 2 s 1 kev 1 ) 10 4 10 3 0.01 0.1 «warm» corona «hot» corona 10 4 10 3 0.01 0.1 1 10 100 Energy (kev)

Model kev 2 (Photons cm 2 s 1 kev 1 ) 10 4 10 3 0.01 0.1 «warm» corona «hot» corona reflection" (xillver) 10 4 10 3 0.01 0.1 1 10 100 Energy (kev)

Model kev 2 (Photons cm 2 s 1 kev 1 ) 10 4 10 3 0.01 0.1 «warm» corona Small BB «hot» corona reflection" (xillver) 10 4 10 3 0.01 0.1 1 10 100 Energy (kev)

Model kev 2 (Photons cm 2 s 1 kev 1 ) 10 4 10 3 0.01 0.1 reddening galactic absorption WA 10 4 10 3 0.01 0.1 1 10 100 Energy (kev)

Results (preliminary) 0.3 τ contours 0.2 L disc,intr L s contours min 0.1 Mkn 509 (Petrucci et al. 2013) 0

Results (preliminary) 0.3 τ contours PG 1116+215 0.2 L disc,intr L s contours min PG 1114+245 0.1 0 Mkn 509 Mkn 279 Mkn 883 The sample agrees with low or even no disc intrinsic emission!

Conclusions A two-corona model (a warm ~1 kev, optically thick τ~10 and a hot ~100 kev, optically thin τ~1) reproduce well the UV-X-ray spectrum of our sample The warm corona explaining the UV-Soft X-ray excess agrees with a slab geometry above the accretion disc MOST of the accretion power is released in the warm corona (illumination? Turbulence?) The warm corona could be the disc upper layers (Janiuk et al. 2001, Czerny et al. 2003, Rozanska et al. 2015)